(71f) On-Board Timing, Memory, and Sensing in Autonomous Cell-Sized Robots Enabled By a Simple Memristor-Based Circuit | AIChE

(71f) On-Board Timing, Memory, and Sensing in Autonomous Cell-Sized Robots Enabled By a Simple Memristor-Based Circuit


Yang, J. F. - Presenter, Massachusetts Institute of Technology
Liu, A. T., Massachusetts Institute of Technology
Zhang, G., Massachusetts Institute of Technology
Brooks, A., Massachusetts Institute of Technology
Koman, V., MIT
Yang, S., Massachusetts Institute of Technology
Liu, P., Massachusetts Institute of Technology
Strano, M. S., Massachusetts Institute of Technology
Tetherless microscale robotic sensors which navigate enclosed spaces and remote locations are becoming a reality. While outstanding research has been done on robotic sensors actuated and supervised externally, progress is slow towards autonomous, intelligent cell-sized agents able to explore an unknown environment all by themselves: The seemingly straightforward path of miniaturizing macroscopic electronics poses an extremely high barrier-to-entry owing to the expertise required to design integrated circuit chips and their hundred-thousand-dollar cost of fabrication.

We demonstrate in this theoretical study that timing, memory, and sensing functionalities can be integrated within a sub-100μm robot with a lean, compact electrical circuit that only comprises simple resistive elements: just an array of memristors in parallel and (chemi)resistors in series. This dramatic reduction in complexity from a foundry-fabricated microchip additionally permits into the playing field a wide range of inexpensive technologies (printing, stamping, and colloidal assembly) as well as materials (polymers and nanomaterials).

Through simulations, we put our microrobotic sensor to the test for four progressively challenging autonomous tasks: (i) keeping track of the elapsed time since departure; (ii) detecting and timestamping a rare event; (iii) collecting and cataloging a series of time-indexed data; and (iv) accessing the stored data for scheduled and stimuli-responsive robotic actions. Through these tasks we establish the suitability of our design for real-world applications: leak detection inside of a remote pipeline, mapping the spatiotemporal analyte profile within an enclosed reactor or in the wild, and autonomous, analyte-responsive delivery of cargos such as insulin analogs. We expect the memristor-based design to become a universal platform for many upcoming autonomous microrobotic sensors in engineering, environmental monitoring, and healthcare as ubiquitous sensor nodes have become an increasingly important focus in the Internet-of-Things era.